Anthropic has introduced new restrictions for Claude Code to prevent users from recreating existing video games with artificial intelligence (AI). The move comes amid growing concerns about AI-assisted game cloning, with popular titles being reverse-engineered and recreated for web browsers and other platforms.
AI tools have increasingly become part of video game development, helping developers create prototypes, write code and experiment with new ideas. However, their growing capabilities have also made it easier for users to reproduce existing games, raising questions about intellectual property, originality and the protection of developers’ work.
The latest restrictions aim to draw a clearer distinction between creating original games and reproducing established titles. Under the updated instructions, Claude Code will no longer help recreate specific existing games, even when users change their names or describe the projects as original creations. Developers can still use the tool to build original games, explore established genres and work with their own designs and assets.
The change was first spotted in a post on X and later confirmed by Anthropic’s Addy Osmani. It reflects growing concerns within the gaming industry about how AI-assisted reverse engineering could affect both major publishers and independent developers.
ANTHROPIC IS CRACKING DOWN ON VIDEO GAME REWRITES 🚨 Claude Code picked up a new rule around Oct 27, sent from Anthropic’s servers under a flag called "tengu_heron_brook". There’s no setting to turn it off. The full rule: “Do build original games, games in a genre, and games from the user’s own designs and assets. Don’t build a recreation of a specific existing game, even if it is renamed or called original.” Is this the end of game rewrites, modding and everything in between?
— chasm (@chasmmmmmmmmmmm) October 30, 2026
AI tools are making popular game recreations easier
Several well-known video games have recently appeared in browser-based versions, reportedly created with assistance from AI coding tools. Among the titles are Halo: Combat Evolved, Grand Theft Auto: Vice City, Call of Duty: Black Ops and The Simpsons: Hit & Run.
These projects demonstrate how AI can reduce the time and technical knowledge traditionally required to recreate complex games. Reverse engineering usually involves examining how an existing game operates, understanding its underlying systems and rebuilding its mechanics and functionality. The process can demand considerable programming experience, particularly when developers attempt to reproduce larger games.
AI coding assistants are changing that process by helping users understand code structures, implement gameplay systems and solve technical problems more quickly. As a result, projects that previously required experienced programmers and lengthy development periods can potentially be completed much faster.
The growing availability of these tools has also drawn attention to games officially restricted to particular platforms. One example is Bloodborne, a PlayStation-exclusive title that has attracted significant interest from the PC gaming community. A community-developed native port project known as bbport has emerged, demonstrating the continued interest in making existing games playable on additional systems.
Bloodborne is running natively on Windows thanks to a community-made PC port, and the latest updates are making it even more impressive. >The project brings the PS4 classic to PC without needing a traditional emulator, with support for Vulkan graphics, resolutions up to 4K, and frame rates of up to 120 FPS. >It also supports AMD FSR 3.1, FSR 4 and NVIDIA DLSS, alongside mods, controller and keyboard/mouse support, and a custom Windows launcher. >The latest v1.6 release adds online features in beta, including messages, bloodstains and wandering ghosts through shadNet. In one test, an RTX 5080 reached around 120 FPS in the area shown.
— Pirat_Nation 🔴 (@Pirat_Nation) October 11, 2026
Although these efforts can generate enthusiasm among gaming communities, they also raise questions about the boundaries between fan projects, technical preservation and unauthorised reproduction. Recreating recognisable games more quickly could create additional challenges for publishers and developers seeking to protect their intellectual property.
Anthropic’s updated restrictions address one part of this issue by limiting Claude Code’s involvement in reproducing identifiable existing titles. However, the distinction between learning from established game mechanics and directly copying a particular game’s implementation remains an important consideration as AI-assisted development becomes more accessible.
Independent developers are concerned about AI game cloning
Concerns about AI-assisted cloning are not limited to major gaming franchises. Independent developers are also taking precautions, with some restricting public access to unfinished projects rather than risk having their work reproduced before its official release.
Digital Sun, the studio behind Moonlighter, recently postponed a planned public Steam playtest for its upcoming game ReVamp. The decision reportedly stemmed from concerns that people could use the publicly available build to reverse-engineer the game and release a competing clone before the original title reaches the market.
Alliance Arts made a similar decision, cancelling its planned Steam Next Fest demo for Drunken Goddess Reflux. Instead of making the demo publicly available, the company opted to organise closed play sessions.
These decisions highlight a growing concern among smaller studios about the risks associated with sharing unfinished games online. Public demos and playtests are traditionally important parts of the development process, allowing studios to introduce their projects to potential players, collect feedback and identify technical problems before launch.
They also give smaller developers a chance to build an audience without relying entirely on expensive advertising campaigns. Players can discover upcoming titles, share their experiences and help generate interest through gaming communities and social media.
However, limiting public access to a demo can reduce these opportunities. Independent studios often have fewer resources than major publishers, making early exposure and community support particularly valuable. The possibility of having a game copied before its official release adds another concern to an already competitive industry.
The situation also raises broader questions about how developers can protect their creative work while still benefiting from AI technology. Although AI can help smaller teams reduce development costs and speed up production, the same technology could potentially be used by others to reproduce their work.
Anthropic draws a line between original development and copying
Anthropic’s decision to introduce restrictions for Claude Code represents an attempt to address concerns surrounding AI-assisted game reproduction. By preventing the tool from recreating specific existing games, the company is setting limits on how its coding assistant can be used within the gaming industry.
The restrictions do not prevent users from developing games inspired by broader genres or experimenting with familiar gameplay concepts. Developers can continue creating original platformers, role-playing games, action titles and other experiences, provided they use their own designs and assets rather than attempting to reproduce identifiable existing games.
This distinction matters because game development has always involved building on established genres and creative ideas. Many successful titles share common mechanics, visual styles and gameplay structures. The challenge is determining when inspiration becomes an attempt to reproduce another developer’s specific work.
The updated instructions also indicate that simply changing a game’s name or presenting a recreation as an original project will not make the request acceptable under Claude Code’s restrictions. This addresses a potential workaround in which users could attempt to bypass safeguards by disguising the identity of the game they want to copy.
However, the practical impact of the restrictions remains uncertain. AI coding tools are only one part of a wider development ecosystem, and similar projects could potentially be created using other software and approaches. The effectiveness of Anthropic’s measures will therefore depend partly on how consistently it applies the restrictions and whether other AI companies introduce comparable safeguards.
For independent developers, the immediate concern remains how to share their work without exposing it to potential copying. For AI companies, the challenge is to support creative experimentation while reducing the risk that their tools will be used to reproduce existing intellectual property.
As AI continues to reshape video game development, the industry will likely face further discussions about ownership, originality and responsible use. Anthropic’s latest restrictions mark a step toward addressing those concerns, although their long-term effectiveness remains to be seen.




